Solid-state dilution of dihydroxybenzophenones with 4,13-diaza-18-crown-6 for photocrystallographic studies

Article abstract of DOI:10.1021/cg201558h

This work forms part of the ongoing drive toward identifying and developing suitable light sensitive substances for photocrystallographic studies. In order to investigate the solid-state dilution of the photoactive dihydroxybenzophenone, X-ray crystal structures of three dihydroxybenzophenone?4,13-diaza-18-crown-6 co-crystals are reported and analyzed. The dihydroxybenzophenone molecules within the co-crystal are compared to those observed in homomolecular dihydroxybenzophenone crystals in terms of their intermolecular contacts, bond geometry and conformation. Molecular volumes and void spaces were calculated using Voronoi-Dirichlet polyhedra and Hirshfeld surface-based space partitioning, demonstrating new ways to represent potential reaction cavities around a photoactive molecule and calculate their packing efficiency. In each case the conditions of solid-state dilution were met. The molecular conformations of the homomolecular environments are retained to varying degrees in the analogous co-crystals. Results show that the co-crystals studied are potentially suitable for photocrystallography. In particular, 2,4-dihydroxybenzophenone molecules in 4,13-diaza-18-crown-6?2(2,4- dihydroxybenzophenone) co-crystals exhibit high structural similarity to their homomolecular analogues suggesting its photochemical properties could be common to both environments.

Full text of DOI:10.1021/cg201558h

Article
pubs.acs.org/crystal
Solid-State Dilution of Dihydroxybenzophenones with 4,13-Diaza-
18-crown-6 for Photocrystallographic Studies
Jacqueline M. Cole,*,†,‡ Paul G. Waddell,†,‡ and Dylan Jayatilaka§
†Cavendish Laboratory, University of Cambridge, J. J. Thomson Avenue, Cambridge, CB3 0HE, U.K.
‡Department of Chemistry, University of New Brunswick, P.O. Box 4400, Fredericton, New Brunswick E3B 5A3, Canada
§School of Biomedical, Biomolecular & Chemical Sciences, University of Western Australia, Crawley, Western Australia, Australia
W
S
* Web-Enhanced Feature * Supporting Information
ABSTRACT: This work forms part of the ongoing drive toward identifying and
developing suitable light sensitive substances for photocrystallographic studies. In
order to investigate the solid-state dilution of the photoactive dihydroxybenzo-
phenone, X-ray crystal structures of three dihydroxybenzophenone·4,13-diaza-18-
crown-6 co-crystals are reported and analyzed. The dihydroxybenzophenone
molecules within the co-crystal are compared to those observed in homomolecular
dihydroxybenzophenone crystals in terms of their intermolecular contacts, bond
geometry and conformation. Molecular volumes and void spaces were calculated
using Voronoi−Dirichlet polyhedra and Hirshfeld surface-based space partitioning,
demonstrating new ways to represent potential reaction cavities around a
photoactive molecule and calculate their packing efficiency. In each case the
conditions of solid-state dilution were met. The molecular conformations of the
homomolecular environments are retained to varying degrees in the analogous co-
crystals. Results show that the co-crystals studied are potentially suitable for photocrystallography. In particular, 2,4-
dihydroxybenzophenone molecules in 4,13-diaza-18-crown-6·2(2,4-dihydroxybenzophenone) co-crystals exhibit high structural
similarity to their homomolecular analogues suggesting its photochemical properties could be common to both environments.
Coincidentally, lowering the concentration of the photo-
active species has a number of additional advantages for the
field of photocrystallography. First, the extinction coefficient, ε,
of the resulting crystal is reduced compared with that of the
homomolecular crystals of the photoactive species. This is a
very important consideration in photocrystallography since ε is
inherently related to the optical penetration depth, 1/μ (μ = εc,
where μ is the absorption coefficient and c the concentration),
which is often the limiting factor in the feasibility of such an
experiment. For example, 1/μ is typically sub-micrometer and
the crystal sample must be smaller than 1/μ in order to ensure
that light passes through the entire crystal; otherwise the
incident light becomes trapped within the crystal resulting in
local heating effects that can literally eat away the crystal or
even cause it to explode.2 This small sample size imposes severe
experimental challenges for even synchrotron-based photo-
crystallography experiments. Therefore, any method that
reduces 1/μ is highly desirable. Second, the light intensity
required to achieve photoexcitation is reduced by lowering the
photoactive concentration. This is because there are less
photoactive species per unit volume that need photoexciting, so
less light energy is exhausted. Finally, the likelihood of triplet−
triplet annihilation decreases as the distance between photo-
INTRODUCTION
■
With the developing field of photocrystallography,1−3 aided by
the advent of third-generation synchrotron sources,4 the
structural determination of the photoactive states of molecules
has become possible.5−8 In recent years, there has been a drive
to identify suitable compounds for structural studies of this
kind. While a plethora of reports exists characterizing the
photophysical and photochemical properties of molecules in
solution, little remains known about their comparative solid-
state optical properties, especially in the periodic state.
Consequently, there has been a considerable effort in designing
a solid-state molecular environment that would allow these
characteristics to be investigated and complement solution-
phase investigations. This way, the myriad of photoactive
compounds whose metastable and time-resolved photophysical
properties have already been characterized in the solution-state
would instantly become accessible to photocrystallographic
characterization; that is, their 4-D (space-time) photoinduced
structural perturbations could be realized.
One approach toward this goal is to “dilute” the photoactive
species in the solid state by creating supramolecular arrays that
incorporate the photoactive species within an inert matrix.
Various methods have been demonstrated that achieve this,
ranging from the co-crystallization of the photoactive species
with an inert molecule to more complicated host−guest
chemistry.9−11
Received: November 25, 2011
Revised: February 13, 2012
Published: March 20, 2012
© 2012 American Chemical Society
2277
dx.doi.org/10.1021/cg201558h | Cryst. Growth Des. 2012, 12, 2277−2287

Crystal Growth & Design
Article
Table 1. Experimental and Refinement Details for the X-ray Structures of 1, 2, and 4−7
1
2
4
5
6
7
chemical formula
C13H10O3
214.21
C13H10O3
214.21
C12H26N2O4
262.35
C38H46N2O10
690.77
C38H46N2O10
690.77
C25H36N2O7
476.56
M
T/K
180
180
180
180
180
180
crystal system
space group
a/Å
monoclinic
P21/n
monoclinic
P21/c
monoclinic
P21/a
triclinic
triclinic
triclinic
P1
̅
P1
̅
P1
̅
7.746(7)
12.245(11)
11.243(10)
90
7.211(3)
19.098(9)
7.513(3)
90
8.879(7)
8.560(6)
10.304(8)
90
8.1924(3)
10.0165(3)
11.7164(5)
106.969(2)
94.936(2)
90.313(2)
915.70(6)
1
9.7905(7)
9.9569(8)
10.8028(9)
63.925(4)
72.701(4)
87.720(4)
898.03(12)
1
8.6935(2)
10.3060(2)
15.8548(2)
90.4906(11)
104.4329(11)
111.0618(9)
1276.19(4)
2
b/Å
c/Å
α (°)
β (°)
109.523(1)
90
92.835(7)
90
109.644(11)
90
γ (°)
U
obs/Å3
1005.1(16)
4
1033.4(8)
4
737.6(10)
2
Z
μ(Mo-Kα) mm−1
Dc/Mg m−3
0.71073
1.416
0.71073
1.377
0.71073
1.181
0.71070
0.71070
0.71070
1.253
1.277
1.240
total reflections
4566
4789
2730
21592
24110
40131
independent reflections (Rint)
max and min transmission
final R, wR2
1943(0.0841)
0.982, 0.982
0.0578, 0.1585
2487(0.0343)
0.989, 0.985
0.0435, 0.1227
2160(0.0531)
0.993, 0.975
0.0799, 0.1995
4176 (0.0597)
0.989, 0.985
0.0434, 0.0818
4129 (0.0658)
0.994, 0.985
0.0501, 0.0963
5848 (0.0477)
0.984, 0.980
0.0387, 0.0548
largest difference: peak, hole/e·Å−3
0.275, −0.259
0.305, −0.215
0.333, −0.460
0.179, −0.183
0.174, −0.255
0.153, −0.203
active species is increased.12 Therefore, lowering the concen-
tration of the photoactive species will extend its triplet lifetime
relative to homomolecular crystals of the same species.
this with direct chemical bonding between a photoactive and
photoinert species where significant energy is required to break
a covalent bond between them; framework disruption from
subsequent chemical reactions, radical formation, or dangling
bonds is likely.
The use of crown ethers to form co-crystals with hydroxyl-
functionalized organic molecules is not unprecedented. As
mentioned previously, 4,4′-dihydroxybenzophenone co-crystal-
lized with 4,13-diaza-18-crown-6 has already been reported,23
as have the similarly functionalized 2,6-dihydroxynaphthalene
and 2,8-dihydroxynaphthalene.24,25 The existence of these co-
crystals gives us confidence that our proposed co-crystals are
attainable.
This paper tests this hypothesis via the synthesis and crystal
structure determination of 2,2′-dihydroxybenzophenone (2,2′-
DHBP), 2,4-dihydroxybenzophenone (2,4-DHBP), and 4,4′-
dihydroxybenzophenone (4,4′-DHBP) co-crystallized with
4,13-diaza-18-crown-6 (DC). The intermolecular interactions
are analyzed as is the relative packing efficiency, which is
probed using two different approaches. The geometry and
stability of each DHBP with respect to DC are assessed. In
order to ascertain the effect of co-crystallization on the
photoactive DHBP species, the molecular geometry and
volumes of the DHBPs were compared with those of their
respective homomolecular DHBP crystal structures.
This paper features the solid-state “dilution” of a series of
photoactive species using co-crystallization techniques. Three
photoactive dihydroxybenzophenones (DHBPs) are co-crystal-
lized with 4,13-diaza-18-crown-6 (DC) which acts as the
photoinert matrix. The interest in dihydroxybenzophenones
stems from the renowned photoactive properties of the parent
compound, benzophenone, which has long been studied for its
ability to absorb and scatter UV light and its role in
photochemical reactions.13−17 The triplet lifetime of benzo-
phenone is observed to be of the order of microseconds.18−20
This lifetime is very attractive from a technical perspective since
microsecond-time-resolved photocrystallography is accessible
via X-ray pulsing from a mechanical X-ray chopper, rather than
necessitating the use of the temporal structure of a synchrotron
particle accelerator. A number of mechanical choppers have
been constructed for such experiments.21,22 As microsecond-
lived triplet states have already been observed in 4,4′-
dihydroxybenzophenone co-crystallized with 4,13-diaza-18-
crown-6,23 this strengthens our confidence in this line of
enquiry.
By functionalizing benzophenone with classical hydrogen-
bond donors (i.e., hydroxyl groups) and embedding them in a
photoinert matrix that contains hydrogen-bond acceptors (i.e.,
DC), we hypothesized that the photoactive structure will be
stabilized by a hydrogen-bonding network, and consequently, a
rigid photoactive structure will result. Achieving such rigidity is
important since the photoactive molecules need to be immobile
within the co-crystallized framework; otherwise they will
manifest molecular disorder, which will render a photo-
crystallography experiment untenable. On the other hand, the
photoactive framework needs to be flexible enough to permit
photoinduced structural perturbations within the molecule
without consequently destroying the framework due to undue
crystal lattice strain. A hydrogen-bonding network is generally
considered to be sufficiently flexible in this regard as these
bonds are weak and primarily electrostatic in nature. Contrast
EXPERIMENTAL SECTION
■
2,2′-DHBP (1), 2,4-DHBP (2), 4,4′-DHBP (3), and DC (4) were
purchased from Sigma-Aldrich and used as supplied without further
purification. Crystals of 1 and 4 suitable for X-ray crystallography were
chosen directly from the purchased sample. Crystals of 2 and 3 were
grown via slow solvent evaporation from methanol. Co-crystal
synthesis required stoichiometric amounts of the DHBP and DC
(1.0 mmol) being stirred under reflux in methanol (ca. 5 mL) until
complete dissolution was achieved. The solution was then cooled to
room temperature, and the products (4,13-diaza-18-crown-6·2(2,2′-
dihydroxybenzophenone) (2,2′-DHBP·DC, 5), 4,13-diaza-18-crown-
6·2(2,4-dihydroxybenzophenone) (2,4-DHBP·DC, 6), 4,13-diaza-18-
crown-6·4,4′-dihydroxybenzophenone (4,4′-DHBP·DC, 7)) crystal-
lized via slow evaporation of the solvent.
2278
dx.doi.org/10.1021/cg201558h | Cryst. Growth Des. 2012, 12, 2277−2287

Crystal Growth & Design
Article
a
Table 2. Selected Bond Distances (Å) and Angles (°) for 1, 2, and 5−7
1
2
5
6
7
C(7)−O(1)
1.251(2)
118.38(12)
120.03(12)
121.58(14)
2.35(4)i
1.2575(14)
117.43(10)
120.19(11)
122.37(10)
1.72(2)ii
1.2340(15)
119.23(12)
121.18(11)
119.48(11)
2.46(2)i
1.2512(8)
117.44(14)
120.71(15)
121.84(13)
2.74(2)iii
1.2334(13)
118.59(10)
119.81(10)
121.56(9)
2.45(2)iii
C(1)−C(7)−O(1)
C(8)−C(7)−O(1)
C(1)−C(7)−C(8)
shortest contact
shortest CO···OC
3.121(2)
4.172(2)
5.777(2)
4.417(2)
6.516(2)
a
Specific contact: (i) C−H···H−C; (ii) O−H···H−C; (iii) H−O···H−C.
X-ray Crystallography. Crystal structure data for 1−4 (Tables 1
and 2) were collected on a Rigaku Saturn 724+ CCD diffractometer
employing a molybdenum X-ray source (λMo,Kα = 0.71073 Å) with
SHINE Optics and an Oxford Cryosystems CryostreamPlus open-flow
N2 cooling device. Cell refinement, data collection, and data reduction
were undertaken via the software Rigaku CrystalClear-SM Expert
2.0.26 Absorption corrections were implemented using ABSCOR.27
Crystal structure data for all DHBP·DCs (5−7) were collected on a
Nonius KappaCCD area-detector diffractometer equipped with a
molybdenum X-ray source (λMo,Kα = 0.71070 Å), an Oxford
Cryosystems Cryostream open-flow N2 cooling device and data
collection software COLLECT;28 cell refinement and data reduction
were conducted using software HKL DENZO and SCALEPACK;29
data were corrected for absorption effects by comparing equivalent
reflections with the program SORTAV.30
The structures were solved by direct methods and refined by full-
matrix least-squares methods on F2 values of all data. Refinements
were performed using SHELXL.31 For all structures, hydrogen atoms
were located from the Fourier difference map of electron density and
refined freely.
Figure 2. Structure of the asymmetric unit of 6 with atomic
displacement ellipsoids drawn at the 50% probability level. Hydrogen
atoms not involved in hydrogen-bonding (dashed lines) are omitted
for clarity.
Crystal structures of the co-crystals of 5−7 (Figures 1, 2, and 3)
were determined at 180 K. While previous room temperature data
equivocal comparison. Given these disorder effects, any comparisons
made between the structures of 3 and 7 should be treated tentatively.
Figure 1. Structure of the asymmetric unit of 5 with atomic
displacement ellipsoids drawn at the 50% probability level. Hydrogen
atoms not involved in hydrogen-bonding (dashed lines) are omitted
for clarity.
exist for 1−4,32−34 their X-ray crystal structures (Figures 4, 5, and 6)
were collected at 180 K for this work to provide a direct comparison to
the DHBP·DC structures. Data for 3 are not reported herein since the
180 K structure is disordered, in common with its room temperature
determination.35 Such disorder compromises the accuracy in bond
geometry to the extent that temperature effects are negligible by
comparison. As such, the previously reported crystal structure for 3 is
used as the 4,4′-DHBP reference for this study since it provides an
Figure 3. Structure of the asymmetric unit of 7 with atomic
displacement ellipsoids drawn at the 50% probability level. Hydrogen
atoms not involved in hydrogen-bonding (dashed lines) are omitted
for clarity.
2279
dx.doi.org/10.1021/cg201558h | Cryst. Growth Des. 2012, 12, 2277−2287

Crystal Growth & Design
Article
Each DHBP molecule in 5 forms significant nonbonded
contacts with nine other DHBP molecules and five DC
molecules, whereas 6 is in contact with seven DHBPs and six
DCs. The shortest DHBP···DHBP contacts are C−H···H−C
interactions in 5 (2.46(2) Å) and an H−O···H−C contact in 6
(2.74(2) Å). Comparing these findings with the supramolecular
environment of their homomolecular DHBP structural
analogues, a similarly short C−H···H−C separation of
2.35(2) Å, as observed in 5, is present in 1. Meanwhile, the
analogous O−H···H−C nonbonded contact in 2 is much
shorter than that in 6, at 1.72(2) Å.
The DHBP molecules in 5 and 6 each interact with the DC
molecule via hydrogen-bonding through one of the DHBP
hydroxyl groups and a nitrogen atom on the DC molecule
(Table 3), forming finite units of two DHBP molecules and one
Figure 4. Structure of the asymmetric unit of 1 with atomic
displacement ellipsoids drawn at the 50% probability level.
a
Table 3. Hydrogen-Bond Geometry (Å, °) for 1, 2, and 5−7
D−H···A
D−H
H···A
D···A
D−H···A
1
O2−H2O···O1
O3−H3O···O1
O3−H3O···O1i
O2−H2O···O1
O3−H3O···O1ii
O2−H2O···N1
O3−H3O···O1
O3−H3O···N1
O2−H2O···O1
O2−H2O···N2
O3−H3O···N1
0.89(3)
0.89(3)
0.89(3)
0.92(2)
0.90(2)
1.05(2)
0.95(2)
1.07(3)
0.94(2)
0.96(2)
1.00(3)
1.87(3)
1.83(3)
2.44(3)
1.72(2)
1.82(2)
1.60(2)
1.70(2)
1.54(3)
1.71(2)
1.78(2)
1.69(3)
2.654(3)
2.603(3)
2.999(3)
2.545(2)
2.726(2)
2.640(1)
2.563(1)
2.600(2)
2.569(2)
2.735(2)
2.671(2)
145(3)
145(3)
122(2)
149(2)
176(2)
173(2)
150(2)
169(3)
152(2)
174(2)
169(2)
2
5
6
7
a
Symmetry codes: (i) 1 − x, −y, 1 − z; (ii) −1 + 5, 1.5 − y, 0.5 + z.
Figure 5. Structure of the asymmetric unit of 2 with atomic
displacement ellipsoids drawn at the 50% probability level.
Figure 6. Structure of the asymmetric unit of 4 with atomic
displacement ellipsoids drawn at the 50% probability level.
Figure 7. View of the finite hydrogen-bonded unit in the structure of
5.
RESULTS AND DISCUSSION
DC molecule (Figures 7 and 8) characterized by the graph set
D22(12) 36 about a crystallographic inversion center. In 5 the
hydroxyl group hydrogen-bond donor lies at the ortho-position
of the DHBP molecule while in 6 the para-hydroxyl group acts
as the hydrogen-bond donor. Intramolecular hydrogen bonds
also exist in 5 and 6 between a hydroxyl group situated in the
ortho-position and the carbonyl oxygen, forming an enol-ring
motif with the graph set S(6). Comparing this hydrogen-
bonding with that of the homomolecular DHBPs 1 and 2, only
■
1. Supramolecular Environment. 2,2′-DHBP and 2,4-
DHBP forms co-crystals with DC (5 and 6) in a 2DHBP/1DC
ratio, while the 4,4′-DHBP analogue (7) exists in a 1DHBP/
1DC ratio. Given these varying ratios, the supramolecular
environments of 5 and 6 are likely to differ substantially from
that of 7. Because of this discrepancy, intermolecular contacts
observed in 5 and 6 are compared together while those for 7
are discussed separately.
2280
dx.doi.org/10.1021/cg201558h | Cryst. Growth Des. 2012, 12, 2277−2287

Crystal Growth & Design
Article
Figure 8. View of the finite hydrogen-bonded unit in the structure of
6.
weak intermolecular hydrogen-bonding is found in the
structure of 1, forming R22(12) and R22(4) motifs about an
inversion center. Hydrogen-bonding where the carbonyl on the
DHBP acts as an acceptor is absent in the structure of 6. This
prevents the formation of chains of connected molecules, akin
to those of graph set C(8) along the crystallographic [102]
direction observed in the structure of 2; in this case, packing is
facilitated by hydrogen bonding involving the ketone oxygen
acting as an acceptor. It is worth noting that 5 contravenes
Etter’s second rule of hydrogen bonding in organic compounds
which states that “six-membered-ring intramolecular hydrogen
bonds form in preference to intermolecular hydrogen-bonds”36
and so 5 represents a rare exception to this rule.
Figure 9. View of the infinite hydrogen-bonding chain along the [111]
direction in the structure of 7.
The 1DHBP/1DC ratio in the structure of 7 arises from the
para-hydroxyl substitution of both benzene rings in 4,4′-DHBP
which enables the involvement of both hydroxyl groups in
hydrogen bonding to a nitrogen on a DC molecule. This
creates infinite chains of hydrogen-bonded molecules along the
[111] direction with the graph set C44(46) (Figure 9). Each
DHBP exhibits nonbonded contacts to four other DHBP and
six DC molecules. There are two crystallographically distinct
DC molecules in 7 corresponding to the different interplanar
angles of the DC rings and the adjacent phenyl rings on the
DHBP. Unlike its homomolecular analogue, 3,35 7 does not
pack in two-dimensional sheets owing to the absence of
hydrogen-bonds bound to acceptors on the DHBP. However,
there are indications of intermolecular π−π interactions
between phenyl groups of the DHBP molecules in adjacent
chains, with centroid···centroid separations of 4.051(5) Å.
2. Conformational Variation of the DHBP Photo-
Congener. The conformation of a benzophenone molecule is
dictated by the interplanar twist angle between the two phenyl
rings.37 In unsubstituted benzophenone, this angle is 54° and
65° respectively for the orthorhombic38 and monoclinic39
polymorphs. The comparative angles in 1 and 5 are 45.57(8)°
and 82.24(4)° respectively. This marked difference can be
attributed to the formation of only one enol ring in 5 compared
with two in 1, rendering a large deviation between the DHBP
conformations in 1 and 5 (root-mean-square deviation (rmsd):
0.884 Å (Figure 10)). There is also substantial variation in
Figure 10. Superposition of the 2,2′-DHBP conformations of 1 (red;
interplanar twist angle = 45.57(8)°) and 5 (yellow; interplanar twist
angle = 82.24(4)°). rmsd = 0.884 Å.
bond angles about C(7) between these structures. In contrast,
the associated twist angles for 2,4-DHBP in 2 and 6 are
51.74(5)° and 57.11(5)°, respectively. A comparison of the two
2,4-DHBP conformers therefore results in a low rmsd (0.127 Å
(Figure 11)). Given that this twist angle can greatly affect the
photophysical properties of these molecules,40−42 we would
expect there to be a discrepancy in the behavior of 2,2′-DHBP
in 1 and 5, while we could expect similar photophysical
properties to be exhibited by the conformationally similar 2,4-
DHBP in 2 and 6.
Meanwhile, the 4,4′-DHBP molecule in 7 exhibits an
interplanar phenyl ring twist angle of 48.83(4)° which is
comparable to either of the two crystallographically analogous
angles in the room temperature structure of 3 (47° and 48°).
3. DHBP Bond Geometry. The aromatic rings in
homomolecular DHBP crystals exhibit characteristic bond
alternation patterns according to the position of hydroxyl
substitution: ortho-quinoidal character in 1 and an average of
both ortho- and para-quinoidal canonical forms in 2 (Figure
2281
dx.doi.org/10.1021/cg201558h | Cryst. Growth Des. 2012, 12, 2277−2287

Crystal Growth & Design
Article
4. Crystal Packing Analysis Using Voronoi−Dirichlet
Polyhedra: Distance-Based Molecular Partitioning. The
extent of solid-state dilution of DHBP molecules, by hosting
them in an inert matrix of DC via co-crystallization, is evaluated
by considering the relative volume ratio of the different
components of the co-crystal. This ratio was calculated by
constructing Voronoi−Dirichlet polyhedra (VDP) for each
molecule using the TOPOS suite of programs (Figures 13 and
14, Table 4).45,46 These volumes can be used to calculate the
percentage composition of the co-crystals in terms of DHBP
and DC (eq 1):
n(VDPDHBP
)
Vol%DHBP
=
× 100%
U
(1)
obs
where VDPDHBP = the VDP volume of the DHBP molecule
within the crystal; Uobs = the unit cell volume from the
crystallographic data; n = the number of DHBP molecules in
the unit cell; Vol%DHBP = the percentage of the volume of the
unit cell which can be assigned to DHBP molecules.
Figure 11. Superposition of the 2,4-DHBP conformations of 2 (red;
interplanar twist angle = 51.74(5)°) and 6 (yellow; interplanar twist
angle = 57.11(5)°). rmsd = 0.127 Å.
Packing efficiency can be assessed by comparing the VDP
volume of the individual molecules in DHBP·DC with those of
the homomolecular crystalline molecules; lower volumes
indicate greater packing efficiency since a residual VDP
represents a manifestation of closer neighboring molecules.
Using the VDP approach may only give an estimate of
molecular volume as it does not account for void space, but as it
accounts for and partitions the total unit cell volume between
the constituent molecules it will serve at least as a robust means
of comparison.
The volume of a molecule of 3 is taken to be the volume of
the unit cell (Uobs), from the previously reported room
temperature analysis of 3, divided by Z.35 In-house crystal
structure data for 3 were collected at 180 K; however the
disorder manifest in the room temperature structure was still
evident. The influence of this disorder on the unit cell volume
is surely greater than the influence of temperature. It was
therefore decided that the previously reported data would be
used for volume calculations. It should be noted that the VDP
volume of a molecule in 3 is likely to be greater than anticipated
owing to disorder of the 4,4′-dihydroxybenzophenone mole-
cules.35,46 %VolDHBP values for 5 and 6 are naturally larger than
that of 7 by simple consequence of the DHBP/DC ratio.
When considering the packing efficiency of 5−7 compared
with the crystal structures of their individual components, we
12). As would be expected the corresponding bond-length
patterns in 5−7 are similar to their homomolecular counter-
parts. Upon excitation we would expect these bond distances to
be altered since π−π* triplet excitation of benzophenone
involves π* electrons which are thought to be delocalized on
the aromatic rings.43
The only marked difference in chemical bonding between
cocrystal and homomolecular variants concerns the carbonyl
group; the C(7)−O(1) distance in DHBP·DC is shorter than
that in their homomolecular analogues. This may reflect the
absence of intermolecular hydrogen-bonds in DHBP·DC where
O(1) is the acceptor. Since vibrational analysis of the
photoexcited state of benzophenone suggests that the CO
bond is severely weakened upon excitation38 and that this bond
is thought to constitute the excitation center,44 the stability of
this bond will be an important consideration. To this end, it is
also important to identify the shortest C−O···O−C distances in
each structure (Table 2). For the most part, this intermolecular
separation is longer in DHBP·DC than in the homomolecular
crystal structures, the most significant difference being between
1 and 5 where the C−O···O−C distances are 3.121(2) Å and
5.777(2) Å respectively. Greater separations between these
excitation centers could reduce the risks of triplet−triplet
annihilation during photocrystallographic studies.
Figure 12. Resonance exhibited by 2,4-DHBP as a result of ortho- and para-hydroxyl substitution.
2282
dx.doi.org/10.1021/cg201558h | Cryst. Growth Des. 2012, 12, 2277−2287

Crystal Growth & Design
Article
Figure 13. View of the VDP of 2,2′-DHBP molecules in the structure of 5.
Figure 14. View of the VDP of the DC molecule in the structure of 5.
Table 4. VDP and Unit Cell volumes (Å3) for 1−7
1
2
3
4
5
6
7
DHBP
DC
251.28
258.35
266.29
255.00
405.7
248.48
401.07
898.03
55.3%
885.49
1.014
246.30
391.79
1276.19
38.6%
368.79
737.6
0%
Uobs
1005.1
100%
1033.4
100%
2130.3
100%
915.70
55.7%
871.35
1.050
%DHBP
Ucalc
1270.16
1.005
U
obs/Ucalc
can compare the volumes of equivalent molecules in the
different systems with lower volumes indicating better packing
efficiency. We can also use VDP volumes to build a calculated
unit cell with volume, Ucalc, by totalling the VDP volumes of the
prerequisite number of DHBP molecules from 1, 2, or 3 and
DC molecules from 4. In 7, two distinct VDP volumes for the
crown ether were observed for the two crystallographic
independent DC molecules; the value used is an average of
the two. By comparing Ucalc to the observed unit cell volume
(Uobs), it can be seen that Ucalc < Uobs for 5−7; hence Uobs/Ucalc
> 1 for all three systems. On the face of it, this implies that the
packing is less efficient overall in 5−7 than in 1−4. While it is
true that the DC molecules have a greater volume in 5−7 than
in 4 and can therefore be said to be packing less efficiently, in 6
and 7 the volumes of the DHBP molecules are lower than that
in 2 and 3. Therefore the DHBP molecules in 6 and 7 can be
said to pack more efficiently with DC than with each other.
However, DC packs more efficiently with itself than with
DHBP to such an extent that the overall unit cell volume in 5−
7 is greater than the calculated value.
These variations in molecular volume between DHBP·DC
and homomolecular crystals can be rationalized by comparing
their hydrogen-bonding motifs. In 5−7, the volume of 4,13-
diaza-18-crown-6 is greater than that observed in 4 and this can
be attributed to the two hydrogen-bonds linked to the nitrogen
atoms on the 4,13-diaza-18-crown-6 in 5−7 where there is no
hydrogen-bonding present in 4. The lower VDP volumes of
DHBP in 6 and 7, compared to those in 2 and 3, can be
rationalized by hydrogen-bonding effects; their extent is lower
in 6 and 7 than in 2 and 3. Meanwhile the weak intermolecular
hydrogen bonds observed in 1 result in a molecular volume
2283
dx.doi.org/10.1021/cg201558h | Cryst. Growth Des. 2012, 12, 2277−2287

Crystal Growth & Design
Article
Table 5. Hirshfeld Surface and Promolecule Density Isosurface Volumes (Å3) for 5−7
5
6
7
DC
DHBP
263.0
DC
DHBP
255.6
DC1
DC2
DHBP
258.2
Uobs.
915.70
898.03
1276.19
Hirshfeld surface (HS) isosurfaces
HS
total HS volume (Å3)
voidHS (Å3)
372.0
369.5
372.0
372.0
1260.4
15.8
898.0
17.7
2.0
880.7
17.3
1.9
voidHS (%)
1.3
Promolecule Density (PD) isosurfaces
PD (iso = 0.0015 au) (Å3)
total PD volume (Å3)
voidPD (Å3)
330.3
89%
238.9
91%
330.5
89%
238.6
93%
330.5
1142.8
133.4
89%
808.1
107.6
807.7
90.3
packing (iso = 0.0015 au) (Å3)
89%
93%
lower than that of the DHBP in 5 wherein more conventional
hydrogen-bonding interactions are observed.
way to characterize void space in a molecule is to locate those
regions with low electron density. Spackman and co-workers
have recommended using an isosurface of the promolecule
density to determine those regions in a unit cell which
correspond to a void;50 the regions exterior to these surfaces
will be good approximation to the low electron-density regions
in the unit cell provided the molecules are uncharged. Here, we
find that the recommended value of 0.0003 au for “permanent
void space” is not appropriate for characterizing the void space
of the structures in this paper, because the volumes associated
with each molecule are too large. We used a value of 0.0015 au
to calculate the isosurfaces (Figures 15, 16, and 17).
5. Crystal Packing Analysis Using Hirshfeld Surfaces:
Electron Density-Based Molecular Partitioning. Although
the VDP analysis gives a notion of the volume of a molecule in
a crystal, there are at least two disadvantages to using it. First,
the VDP volume is independent of the types of atoms that
compose the molecule. Only the atom positions are relevant,
and its size and charge are irrelevant. Second, the VDP
partitioning of space is exhaustive. Thus there is no notion of
“void” or “space” in the structure using a VDP analysis. In view
of these issues, it is worthwhile to consider alternative ways to
quantify notions of void space and packing efficiency.
The Hirshfeld surface47 is similar to the VDP partitioning
scheme in that space partitioning is based on a concept of
“ownership” by a molecule. However, rather than using a
distance criteria to decide ownership, the Hirshfeld partitioning
uses an electron density criteria. In fact, the Hirshfeld surface is
defined as the region where 50% of the promolecular electron
density comes from the enclosed molecule, and 50% comes
from all surrounding molecules (the promolecular electron
density is an approximation to the actual electron density
comprised of the sum of isolated spherical atomic electron
densities; the Hirshfeld partitioning would be essentially
identical to the VDP analysis if all the atoms were assigned
the same electron density).
Although the relative sizes of each atom are taken into
account when Hirshfeld surfaces are used, the actual charge
density itself is not taken into account. Furthermore, the
Hirshfeld surface partitioning is not exhaustive: there are parts
of the crystal structure which are not associated with any
molecule, and Hirshfeld surface “void space” is an important
component of the packing density (eq 2):
Figure 15. Void surfaces (0.0015 au) for 5. The blue rendering
represents volumes of low electron density in one unit cell. The
boundaries of the cell are denoted by the multicolored capping points
at the boundaries of each face. A dynamic version of the figure in AVI
format is available.
U
− ∑ VolHS
∑ VolHS
obs
%HSvoid
=
(2)
Data for the voids in the crystal structures are presented in
Table 5. At a density d = 0.0015 au it quite clear that 6 has the
lowest void space (90.3 Å3), close to 5 (107.6 Å3) and
significantly less than 7 (133.4 Å3). The latter is because there
are two crown ethers in the asymmetric unit for 7 but only one
for the other two structures.
Although the promolecule density method easily classifies
void space, it is not able to quantify the amount of void
associated with a particular molecule. One of the main aims of
this work is to provide some estimate of the “space” available
for each molecule to maneuver within the crystal lattice when it
where %HSvoid is the percentage of Hirshfeld surface void
space; Uobs is the observed unit cell volume; VolHS is the
Hirshfeld surface volume.
It can be noticed from Table 5 that the Hirshfeld surface
volume, calculated using CrystalExplorer 2.1,48 is nearly
constant with 7 having a slightly larger volume. This structure
also has the lowest percentage (1.3%) of Hirshfeld surface void
space, while 5 and 6 have about the same (2%). However, it is
clear that the Hirshfeld-surface volume analysis alone is
uninformative in an absolute, quantitative sense.49 The best
2284
dx.doi.org/10.1021/cg201558h | Cryst. Growth Des. 2012, 12, 2277−2287

Crystal Growth & Design
Article
Table 5 shows the packing efficiencies calculated using this
method at a density d = 0.0015. For the crown ether, the
packing efficiency for all structures is identical (89%). On the
other hand 6 has a slightly higher packing efficiency (93%)
compared to 5 (91%). This is consistent with it having the
lowest void space. However, the packing efficiency of 6 is the
same as 7 (93%).
6. Suitability for Photocrystallography. The promole-
cule density approach for identifying void space within these
structures reveals that there are significant cavities within the
structures as a result of the molecular packing (Figures 15, 16,
and 17). The blue rendering in these figures represents areas of
low electron density within one unit cell into which a molecule
could more easily maneuver when adjusting to the effects of
photoexcitation. The figures are oriented in such a way as to
readily observe the void space surrounding the carbonyl group;
since previous spectroscopic studies have shown this bond to
weaken upon photo excitation,43 such void space is required to
enable the associated geometric perturbation. The weakening of
the carbonyl moiety may also impact upon the conjugated
phenyl rings as the distribution of electron density within them
may change to some extent upon excitation. Hence these rings
may also require room to maneuver when photoexcitation
occurs.
Considering the void space within the structure of 5 (Figure
15), the carbonyl appears to be perpendicular to much of the
void space due to its parallel orientation relative to the phenyl
group which is also parallel to the void space. As a consequence
of this 5 appears to be the most constrained of the three co-
crystal structures in terms of the reaction cavity of its carbonyl
group.
Figure 16. Void surfaces (0.0015 au) for 6. The blue rendering
represents volumes of low electron density in one unit cell. The
boundaries of the cell are denoted by the multicolored capping points
at the boundaries of each face. A dynamic version of the figure in AVI
format is available.
The carbonyl group on the DHBP in 6 is significantly out of
the plane and directed toward void space that intersects the
DHBP and DC molecules (Figure 16). Interestingly, the
hydroxyl-substituted phenyl ring is also surrounded by a
substantial amount of void space, should this ring need to move
to some degree.
In the structure of 7, the carbonyl group is directed between
two pillars of void space giving it room to maneuver (Figure
17). However the reaction cavity can be said to be constrained
in a purely translational manner as the carbonyl oxygen is not
directed toward any void space which would seem to hinder
any potential elongation. Its location within a channel of void
space may be satisfactory as the whole molecule may maneuver
into the large amounts of void space that surround the
molecule, particularly the two phenyl rings. While it is unlikely
to affect the photoactivity, the center of the DC molecule in all
three examples also contains a large volume of void space as
might be expected from such a large ring molecule with a space-
filled center.
Comparing the contact between photoactive molecules in
resorcinarene-based systems incorporating benzophenone9−12
to those of DHBP in the DHBP·DC co-crystals, it can be seen
that in cases where cavities contain benzophenone dimers, this
intermolecular contact can be said to be greater, whereas it is
decreased in cases where benzophenone occurs as a monomer.
Some of these co-crystals may be well suited to the
incorporation of DHBP due to the presence of hydrogen-
bond acceptors on the framework molecules (e.g., C-
methylcalix[4]resorcinarene·2(trans-1,2-bis(4-pyridyl)-
ethylene)).12 Were this to be achieved, the DHBP molecules
are less likely to be disordered, as is observed in some cases in
Figure 17. Void surfaces (0.0015 au) for 7. The blue rendering
represents volumes of low electron density in one unit cell. The
boundaries of the cell are denoted by the multicolored capping points
at the boundaries of each face. A dynamic version of the figure in AVI
format is available.
undergoes a photochemical process. To quantify this space, we
defined the molecular packing efficiency by the volume of a
promolecule density isosurface of isovalue d, divided by the
corresponding Hirshfeld surface volume (eq 3):
Vold
VolHS
%Packing Efficiencyd =
(3)
2285
dx.doi.org/10.1021/cg201558h | Cryst. Growth Des. 2012, 12, 2277−2287

Crystal Growth & Design
Article
benzophenone when incorporated into these frameworks,9,11,12
due to the herein proposed intermolecular hydrogen bonding.
The prediction power of intermolecular hydrogen-bonding is
also exemplified in this work. The nature of the DHBP
molecules in their DC co-crystal environments indicate how
accurate a photophysical and photochemical property pre-
diction of homomolecular DHBP crystals could be achieved
through photocrystallographic studies of these co-crystals. It is
not unreasonable to say that studies of DHBP·DC will give a
very accurate picture of the homomolecular systems wherever
the DHBPs are similar in both co-crystalline and homomo-
lecular crystals with respect to the conformation of the DHBP
molecule and the numbers of hydrogen-bond acceptors and
donors. The DHBP molecules in 2 and 6 are observed to be
similar in terms of conformation, and given the importance of
twist angle in determining photophysical properties, this a good
indication that the two will exhibit similar properties.40−42
Their hydrogen-bonding environments are also similar; the
DHBP in 6 donates two hydrogen bonds and accepts one,
identical to the DHBP in 2 but for the one extra intermolecular
hydrogen bond accepted by the carbonyl oxygen. Also worth
noting is that the difference between the C(7)−O(1) bond
distances in 2 and 6 is the lowest of the three systems. In terms
of photophysical properties it can be theorized that of these
three DHBP·DC structures, 6 is the most likely to have similar
properties to its homomolecular analogue. Given that 6 exhibits
longer shortest-contact and carbonyl−carbonyl distances
compared to 2 we expect that the threat of triplet−triplet
annihilation in 6 will be much lower than in 2.23
characterization given the historically high profile of the parent
compound benzophenone within photochemistry.
ASSOCIATED CONTENT
■
S
* Supporting Information
Crystallographic information for 1, 2, and 4−7 are collected
and reported in .cif format. This information is available free of
charge via the Internet at http://pubs.acs.org/.
W
* Web-Enhanced Feature
Dynamic versions of Figures 15−17 are available in AVI format.
AUTHOR INFORMATION
■
Corresponding Author
*E-mail: jmc61@cam.ac.uk.
Notes
The authors declare no competing financial interest.
ACKNOWLEDGMENTS
■
J.M.C. thanks the Royal Society for a University Research
Fellowship, the University of New Brunswick for the UNB
Vice-Chancellor’s Research Chair, and NSERC for the
Discovery Grant 355708 (for P.G.W.). The University
Chemical Laboratory, Cambridge, is also acknowledged for
part-provision of X-ray diffraction and synthetic laboratory
facilities, via EPSRC Grant GR/R07097. Paul Raithby
(University of Bath) is also thanked for helpful discussions.
Comparing 1 to 5 it was shown that each DHBP in 1 donates
to four hydrogen bonds and accepts three where in 5 each
DHBP donates to two and accepts one. This discrepancy in
hydrogen-bonding environment coupled with the drastically
different conformations of DHBP in the two structures suggests
that the 1 and 5 may have different photophysical properties.
The same can be said for 3 and 7 where there may be
similarities in terms of conformation but there are twice as
many hydrogen-bonds associated with one DHBP molecule in
3 than there are in 7.
REFERENCES
■
(1) Cole, J. M. Acta Crystallogr., Sect. A 2008, A64, 259.
(2) Cole, J. M. Chem. Soc. Rev. 2004, 33 (8), 501.
(3) Coppens, P.; Fomitchev, D. V.; Carducci, M. D.; Culp, K. Dalton
Trans. 1998, 865.
(4) Coppens, P. Synchrotron Radiat. News 1997, 10, 26.
(5) Kovalevsky, A. Y.; Bagley, K. A.; Coppens, P. J. Am. Chem. Soc.
2002, 124, 9241.
(6) Kovalevsky, A. Y.; Bagley, K. A.; Cole, J. M.; Coppens, P. Inorg.
Chem. 2003, 42, 140.
Finally, considering macroscopic level optical effects, we can
hypothesize that the extinction coefficient will be lower in the
co-crystals than that of the homomolecular crystals owing to
the solid-state dilution of the photoactive molecule; a lower
intensity of light should therefore be adequate for photo-
crystallography experiments. We can also expect the possibility
of triplet−triplet annihilation to have been reduced, although
the possibility remains as there are still contacts between DHBP
VDP. The lack of disorder in the DHBP·DC systems is another
advantage with respect to the feasibility of photocrystallography
experiments.
(7) Bowes, K. F.; Cole, J. M.; Husheer, S. L. G.; Raithby, P. R.;
Savarese, T. L.; Sparkes, H. A.; Teat, S. J.; Warren, J. E. Chem.Commun.
2006, 2448.
(8) Phillips, A. E.; Cole, J. M.; d’Almeida, T.; Low, K. S. Phys. Rev. B
2010, 82 (15), 155118.
(9) Ma, B.-Q.; Zhang, Y.; Coppens, P. Cryst. Growth Des. 2001, 1 (4),
271.
(10) Ma, B.-Q.; Coppens, P. Chem. Commun. 2003, 412.
(11) Ma, B.-Q.; Coppens, P. Cryst. Growth Des. 2004, 4 (6), 1377.
(12) Coppens, P.; Ma, B.-Q.; Gerlits, O.; Zhang, Y.; Kulshrestha, P.
CrystEngComm 2002, 4 (54), 302.
(13) Pitts, J. N.; Letsinger, R. L.; Taylor, R. P.; Patterson, J. M.;
Recktenwald, G.; Martin, R. B. J. Am. Chem. Soc. 1959, 81 (5), 1068.
(14) Cohen, S. G.; Parola, A.; Parsons, G. H., Jr. Chem. Rev. 1973, 73
(2), 141.
(15) Scaiano, J. C. J. Photochem. 1973, 2, 81.
(16) Bisby, R. H.; Johnson, S. A.; Parker, A. W. Free Radical Biol.
Med. 1996, 20 (3), 411.
(17) Favaro, G. Chem. Phys. Lett. 1973, 21 (2), 401.
(18) Schuster, D. I.; T. Weil, M. Chem. Commun. 1971, 1212.
(19) Tahara, T.; Hamaguchi, H.; Tasumi, M. J. Phys. Chem. 1987, 91,
5875.
(20) Husheer, S. L. G.; Cole, J. M.; d’Almeida, T.; Teat, S. J. Rev. Sci.
Instrum. 2010, 81, 043905.
CONCLUSION
■
In conclusion, solid-state dilution via co-crystallisation presents
a very attractive sample preparation method for the study of
solid-state photoexcited states. Such experiments are beginning
to be able to quantify the 4-D (space-time) electronic structure
of these states via photocrystallography. Solid-state character-
ization at such a detailed level of atomic resolution is
unavailable by any other means; yet, it is increasingly important
due to the increasing prevalence of solid-state device design in
optoelectronic applications. The study of DHBP derivatives in
this context is very relevant to this developing field of structural
(21) Coppens, P.; Fomitchev, D. V.; Carducci, M. D.; Culp, K. J.
Chem. Soc., Dalton Trans. 1998, 865.
2286
dx.doi.org/10.1021/cg201558h | Cryst. Growth Des. 2012, 12, 2277−2287

Crystal Growth & Design
Article
(22) Coppens, P.; Benedict, J.; Messerschmidt, M.; Novozhilova, I.;
Graber, T.; Chen, Y.-S.; Voronstov, I.; Scheins, S.; Zheng, S.-L. Acta
Crystallogr., Sect. A 2010, 66 (2), 179.
(23) Zhang, Y.; Wu, G.; Wenner, B. R.; Bright, F. V.; Coppens, P.
Cryst. Eng. 1999, 2 (1), 1.
(24) Watson, W. H.; Nagl, A.; Eduok, E. Acta Crystallogr., Sect. C
1989, C45, 303.
(25) Watson, W. H.; Nagl, A.; Kashyap, R. P. Acta Crystallogr., Sect. C
1991, C47, 800.
(26) CrystalClear-SM Expert and CrystalStructure; Rigaku: The
Woodlands, Texas, USA, 2009.
(27) Higashi, T. ABSCOR; Rigaku Corporation: Tokyo, Japan, 1995.
(28) COLLECT, Data Collection Software; Nonius BV.: Delft, The
Netherlands, 1998.
(29) Otwinowski, Z.; Minor, W. Method. Enzymol. 1997, 276, 307.
(30) Blessing, R. H. J. Appl. Crystallogr. 1997, 30, 421.
(31) Sheldrick, G. M. Acta Crystallogr., Sect. A: Found. Crystallogr
2008, A64, 112.
(32) Liebich, B. W. Acta Crystallogr., Sect. B: Struct. Sci. 1979, B35,
1186.
(33) Schlemper, E. O. Acta Crystallogr., Sect. B: Struct. Sci. 1982, B38,
1619.
(34) Dokurno, P.; Trokowski, R.; Kosc
́
iuszko-Panek, B.; Ossowaski,
T.; Konitz, A.; Błazejowski, J. Z. Kristallogr. 1997, 212, 362.
̇
(35) Ferguson, G.; Glidewell, C. Acta Crystallogr., Sect. C 1996, C52,
3057.
(36) Etter, M. C. Acc. Chem. Res. 1990, 23 (4), 121.
(37) Cox, P. J.; Kechagias, D.; Kelly, O. Acta Crystallogr., Sect. B 2008,
B64, 206.
(38) Moncol, J. Coppens, P. Private communication (refcode
BPHEN012), CCDC, Cambridge, England, 2004.
(39) Kutzke, H.; Klapper, H.; Hammond, R. B.; Roberts, K. J. Acta
Crystallogr., Sect. B 2000, B56, 486.
(40) Singh, A. K.; Palit, D. K.; Mittal, J. P. Res. Chem. Intermed. 2001,
27, 125.
(41) Mondal, J. A.; Ghosh, H. N.; Ghanty, T. K.; Mukherjee, T.;
Palit, D. K. J. Chem. Phys. A 2006, 110, 3432.
(42) Pal, T.; Paul, M.; Ghosh, S. J. Mol. Struct. (THEOCHEM) 2008,
860, 8.
(43) Hoffmann, R.; Swenson, J. J. Phys. Chem. 1970, 74, 415.
(44) Dym, S.; Hochstrasser, R. M.; Schafer, M. J. Chem. Phys. 1968,
48, 646.
(45) Blatov, V. A.; Shevchenko, A. P.; Serezhkin, V. N. Acta
Crystallogr., Sect. A 1995, A51, 909.
(46) Baburin, I. A.; Blatov, V. A. Acta Crystallogr., Sect. B 2004, B60,
447.
(47) Hirshfeld, F. L. Theor. Chim. Acta 1977, 129.
(48) Wollf, S. K.; Grimwood, D. J.; McKinnon, J. J.; Jayatilaka, D.;
Spackman, M. A. CrystalExplorer 2.1; University of Western Australia:
Perth, Australia, 2007.
(49) Spackman, M. A.; Jayatilaka, D. CrystEngComm 2009, 11, 19.
(50) Turner, M. J.; McKinnon, J. J.; Jayatilaka, D.; Spackman, M. A.
CrystEngComm 2011, 13, 1804.
2287
dx.doi.org/10.1021/cg201558h | Cryst. Growth Des. 2012, 12, 2277−2287